Bad Medicine

Wouldn’t it suck if you were compelled to take medicine that you don’t think is helpful? Your bone-headed neighbors could impose bad medicine on you by majority vote. If you feel you are well informed and have very good reason to believe your neighbors are advocating snake-oil, they are under no obligation to listen to your arguments. Indeed, they create an atmosphere in which it is considered “rude” to discuss your objections to compulsory medicine in polite company; some in fact create an atmosphere in which doubting compulsory blood-letting (or needle-pricking or homeopathy) leads to shrill denunciation. Voters don’t directly pick the medical regimen, but rather pick individuals who promise their own favorite folk remedies. These individuals compete for the attention of intellectually lazy voters in a WWE-style soap opera and win more on the force of personality than on the scientific merits of their medicine. This is a world where ignorant people can inject you with bad medicine against your will, and to add insult to injury you get saddled with the bill.

And by “wouldn’t it suck” I of course mean “Doesn’t it suck,” because I have actually described the world we live in. Government policy is mostly bad medicine. Many of the things government does are demonstrably ineffective or even counterproductive. But the public asks for bad medicine and mostly gets what it wants. Gun control is snake oil for the violence problem. Minimum wage laws are homeopathy for the poverty problem. Drug prohibition is acupuncture for the addiction problem. Taxes on capital are aroma therapy for the inequality “problem.” Immigration restrictions are flagellations for the unemployment problem. Keynesian “stimulus” is blood-letting for economic recessions. There is left-wing-populist snake oil and right-wing-populist snake oil, much of it overlapping.

Government policy is medicine for various social problems, and most of it doesn’t work. We can evaluate government policy the same way we evaluate medicine for effectiveness: regression analysis, time series, randomized experiment, instrumental variables, (insert your favorite statistical method here), etc. We somehow compare a “treated” to an “untreated” population, we measure the difference in outcomes, we determine how much of the difference in outcome is due to the treatment, we quantify the uncertainty in our estimate (is it “statistically significant” or not?), and we decide whether the treatment is cost-justified or not. For most of what the government does, you either can’t find any effect at all or you find the opposite of the intended effect. It’s hard to even “move the needle” in the targeted direction. For many policies, there is a small positive effect, but it is purchased at such an enormous cost that the policy can only be called a failure. For other policies, the point estimate for net benefits (total benefits – total costs) may be positive, but the uncertainty around the estimate is so huge that you have no idea whether it’s a net benefit or not. It shouldn’t be terribly controversial to call for the elimination of government programs that are “bad medicine” in one of the ways articulated above. But even if medicine is provably, uncontrovertibly effective *as well as* being cost effective, should you then be able to force it on an unwilling patient?

Most people would not hold down an unwilling victim and inject them with unwanted medicine. Even people who despise anti-vaxers mostly don’t think we should forcibly vaccinate them or even their children. Of course there are some decisions that affect the entire society and thus must be shared. We can’t all have our own “nuclear deterrence” policy, and I can’t have a different pollution standard on my plot of land than my neighbor’s plot mere feet away. Likewise there can only be one “monetary policy” governing the dollar. But I can set my own personal minimum wage, as in “I will not work for less than $X/hour.” This has the advantage that I can change my policy if I’ve made a terrible mistake of setting X too high, which would preclude my finding a job, or too low, which would leave me undercompensated for my skills and experience. I can set my own “tolerance toward immigrants” policy. If I am culturally benighted and find immigrants of certain nationalities distasteful I can mostly avoid them. I can set my own “drug prohibition” policy; perhaps my homeowners association or condo or apartment complex wishes to enforce a “no drugs” policy. That’s fine; the people who want a drug free environment can pay the costs of maintaining it. People who want drugs can use them without being beaten up, and they are spared the indignity of paying for the government policies that abuse them. Those organizations attempting to enforce local drug prohibition might be faced with the realization that enforcement is too costly and too destructive of personal privacy. This is a feature, not a bug, of having private rather than public policy. The discovery process that identifies good and bad policy is allowed to operate in the private sphere; the same discovery process either does not take place *at all* or takes place very slowly and clumsily in the public sphere.

There you have two good reasons for a presumption *against* government coercion. One is strictly moral: the thing that government does is almost universally recognized as immoral in a slightly different context. The other is practical: doing things coercively shuts off the discovery process, whereby individuals who make mistakes can adjust their behavior.

Funding a College Education

Some basic concepts are missing in the public discussion of college education financing. There are individuals who should not go to college, and we should stop setting them up for failure. They hate school. They lack the combination of conscientiousness and intelligence that is required to benefit from a college education. They simply won’t finish, or they will finish with a garbage degree and no useful job skills. We can predict fairly well who will and will not finish college based on their academic performance and test scores collected by the end of high school; this information should certainly be used to sort high- and low-risk students. It would be merciful to spare a likely future drop-out of the massive debt accrued from a failed education. A college education is an *investment.* For those students who make good on their education, the payoff is very big and the investment is worth it. For those students who fail to make good use of their time in college, the investment has a negative return. Indeed, the investment has a negative return *regardless of who foots the bill.*

This point is what annoys me most about the debate over public financing of college education. So much time is wasted and ink spilt over college “affordability.” Affordability isn’t a problem if you finish your degree in a reasonable time frame and graduate with a marketable degree. The “wage premium”, the increase in your future earnings due to finishing college, is enormous for these students. They can finance the high cost of college with their future earnings. (Crudely, college graduates earn ~80% more than high school graduates. Controlling for the fact that students who finish college are initially more talented than those who don’t (ability bias), you get something in the 30-60% range, depending on the student’s major. See here .)

Suppose a college education increases your income by $10,000/year forever. At a 5% discount rate, that is worth $200,000 dollars. (Hint for people who don’t like present value calculations: $200,000 earning 5% interest annually will get you $10,000 a year forever.) It’s worth paying up to $200,000 for this education. Of course, that “payment” includes more than just the tuition payments; it includes the opportunity cost of not working for 4+ years, the sometimes agonizing, long hours of study, and many other costs. Then again, if we think of college as a consumption good, like going to a country club for 4 years, these costs are “offset” by the benefits of that consumption. But I digress. The point is there is a huge value-added, and a huge wage premium from which one can finance a college education. Going $50k - $100k into debt is easily worth it if it returns $10k a year.

Suppose on the other hand that college only increases your income by a few hundred dollars a year, or doesn’t increase your future earnings *at all*. These scenarios are likely for someone who gets a bullshit degree or who doesn’t finish their education. If college only increases your earnings by $1,000/year, then (again at 5% interest rate) this investment only worth $20,000. If four years of tuition costs $50,000, then the problem is *not* affordability. The problem is that the investment just plain isn’t worth making. Changing who pays doesn’t change that. If you do insist on spending $50,000 to help this person, it would be far more merciful to just give him the money, or an annuity of comparable value. He won’t be set up for failure, he won’t drop out of college with massive debt, he won’t feel the sting of failing something he intended to finish.

“College affordability” is a red herring. It is true that sometimes considerations of affordability cause a person to decide against a college education. That is a feature, not a bug. “Unaffordable” usually actually means “the investment doesn’t pay off, *regardless of who pays for it*.” Such investments shouldn’t be made with public money.

There are various rationales for the government to subsidize or finance higher education, and there are obvious drawbacks to such a system. I’ll deal briefly with a few of them, although honestly each of these deserves a book-length treatment.

Maybe financial markets are inefficient and so they don’t bother to finance college, even though objectively it’s a good investment. Or maybe they will only do so at extremely onerous interest rates. I don’t find these stories terribly compelling. Financial markets are always looking for ways to create value (of course, so they can capture some of it). In the above example, we had a $50,000 investment with a $200,000 payoff (I think these are somewhat realistic numbers). That’s an enormous surplus, so it’s just implausible that no investors would be willing to finance it. Lenders are always trying to undercut each other to attract good credit risks, so we should expect that the interest payments on such a loan wouldn’t be overly burdensome. Lenders would have various criteria for who is or isn’t likely to finish college, who is a good credit risk *assuming* they finish, what is the likely wage premium for this individual, etc. They would be part of the sorting mechanism that determines who does and who does not go to college. This does not put us back in a world where only the rich can go to college. A poor-but-talented individual will look like a good investment to several lenders. At any rate, government funding could limit itself to scholarship for such poor-but-talented students, or subsidize the interest repayment for those talented individuals. What it should *not* do is subsidize education so thoroughly that we get a lot of negative-payoff investments as described above.

There are also externalities (costs or benefits to third parties) to consider. First, the signaling theory of education. If a college degree is mostly “signaling”, then your degree raises your status relative to everyone else, but doesn’t make you any more productive. All the time and resources spend educating you are, from society’s point of view, a waste. Subsidizing education would be like subsidizing peacocks to grow bigger, more extravagant tail feathers. The peacock’s colorful fan has no practical use, but it does allow peahens to discriminate good and bad potential mates. We wouldn’t want to subsidize this already-wasteful arms-race, which is closely analogous to the human race for better credentials. In this sense, subsidizing college education is like subsidizing the purchase of nicer interview suits. You could even show that the job applicants with “subsidized” interview suits did better than those with “unsubsidized” interview suits, but it would be a signaling effect, not a real boost to worker productivity. Few people think that college education is actually 100% signaling, but some economists who study this topic say the college wage premium is up to 80% signaling and only 20% skill-building.

Supposedly there are also positive externalities from education, but the discussion of this is very muddled. Supposedly getting an education makes you a “more productive member of society.” Well, you capture that extra productivity in your personal income, so it’s difficult to argue that this is truly an externality. A related argument is that being better educated makes you a better voter, and thus a better educated society will pick smarter policies. This is plausible if you compare levels of education in dysfunctional countries to modern developed democracies. But I think it’s overblown, and it’s certainly hard to tell a causation story here. Certainly most of the highly educated individuals on my Facebook feed aren’t competent to do a cost-benefit analysis. Most of them would probably even fail to articulate the relevant trade-offs for any given policy. The positive externality story is mostly fluff. People often support policies they like by listing good things about it while failing to quantify those things *at all*; that’s exactly what’s happening here. It’s hard to say whether accounting for externalities should make us want to subsidize or tax college educations, but there’s at least a plausible argument that it should be a net tax.

The Social Cost of Drug Use is Grossly Exaggerated

Are there really “externalities” from drug use? When a drug user does harm to a third party, is it correct to say that drug use was the “cause” of that harm?

I want to cast some doubt on the notion that drugs are responsible for significant externalities, and I’d like to cast even greater doubt on the notion that drug prohibition is a reasonable solution to this problem.

If crime is penalized, is it still an “externality?”

There are many mechanisms by which drug use supposedly leads to social harms, but most can be summarized as “drugs dominate the will and turn the user into a criminal or a sociopath.” Supposedly, the drug user loses his mind and becomes irrational and violent. Or perhaps the user’s compulsion to acquire drugs becomes so strong that s/he resorts to economic crimes to pay for the expensive habit. Or perhaps the user drives a vehicle while intoxicated. Or she neglects her children. Or he will disappoint his parents or social group.

No doubt these are all very real harms and it is worth trying to avoid or ameliorate them. However, 1) these are overblown and unlikely consequences for any given drug user and 2) these behaviors are already very heavily penalized! Unpenalized theft would be an “externality,” in some bizarre world where society failed to criminalize theft. The thief gains and a third party (the victim) is harmed. When you appropriately criminalize and penalize theft, those costs are “internalized.” That is, the thief, facing the threat of a penalty that is sufficiently large, faces the full social cost of his action. Some thieves get away with their theft, but the ones who get caught are penalized enough to set a strong example to other would-be thieves. The same argument goes for any other crime: murder, assault, rape, fraud, DUI, child neglect. If the problem is that these behaviors aren’t sufficiently criminalized, then the solution is to address these problems directly. If it is necessary to devote more police resources to solving these problems (likely to work, IMHO) or increase the severity of penalties for those who get caught (unlikely to work, IMHO), then we should explore these solutions. It is extremely inefficient to target drug use, which has at best a tenuous influence on a person’s criminality.

Assuming the penalties for these crimes have been appropriately set, drug-induced criminal behavior is *not* an externality. The potential drug user already faces the threat of a penalty, and moderates their drug use accordingly. It is the same as a potential CO2 emitter facing a perfect carbon tax while choosing how much to emit. The external cost is already internalized, so it ceases to be an externality. That’s not to say there is no harm. When the air is polluted or there is an unnecessary death or beating, somebody surely is harmed. It's just that the polluter (or criminal) pays the social cost of the harm. Suppose we had an appropriately high carbon tax, and suppose drug use induced people toward excessive CO2 emissions. We wouldn’t consider it an “externality” problem or a drug problem. We might adjust the size of the tax to address the problem directly, which is probably what we should do for drug-induced crime.

 If the problem is that people actually don’t understand the causal link, that “crack will turn you into a murderer/thief/assailant with an X/Y/Z% probability”, then the solution is some kind of ad campaign with credible information about these risks. The solution is not a blanket criminalization of drug use.  The potential user with accurate information will grind through the logic: “This habit will turn me into a criminal with probability X, which will almost surely land me in prison for Y years…”(No, he won’t literally do this, but will do so in an approximate fashion.) In this sense drug-induced criminal behavior is already criminalized and the penalty is already built into the cost of doing drugs. It’s already part of the price the user faces. If you object that drug users don’t have the foresight to consider the cost to himself of his future criminal behavior, then you should seriously doubt the entire enterprise of drug prohibition in the first place!  Come up with whatever assumptions you prefer about drug users, their degree of rationality, their behavior, their response to costs, etc.; these assumptions have to be consistent. They can’t contradict each other, which (I think) is what some very confused, very sloppy drug warriors have done. You don’t get to assume that drug users miscalculate the pharmacological risks of drug use but correctly estimate the legal risks. Either their talents for risk-estimation are well-calibrated, or they are not.

Perhaps the large social harms from drug use are due to things that aren’t criminalized. Things like neglect of one’s family or career are either not criminalized or are under-policed. If anyone is proposing we criminalize these behaviors, please suggest a proposal that makes sense, one that doesn’t further destroy the family. Clearly throwing a borderline negligent parent into prison has some harmful ramifications for his/her children, and throwing the same parent in prison on a drug charge has the same effect. If we’re not going to criminalize Behavior A, it makes little sense to criminalize Behavior B which (again tenuously) causes Behavior A.

I’ll say in passing a couple of other things that are relevant to this discussion. Originally this post contained two long sections developing these points, but one big idea per post is enough. First, it’s questionable whether drugs truly “cause” criminality (err..non-drug criminality anyway, the kind with an obvious victim). Certainly they don’t for most people who use them. Indeed for most users, drugs probably have no tendency whatsoever to lead to criminal behavior, and in others it possibly interacts with a pre-existing predilection toward criminality. Some drugs for some people likely *inhibit* criminal tendencies (think a life-changing psychedelic vision quest, which many people have experienced). It makes little sense to summarize this complex causal link as “drugs cause criminal behavior.” It irritates me, the facile way that people assume this causal link exists and is strong enough to justify harsh legal penalties. It’s a conclusion that needs to be argued for, using all the standard tools that social scientists use to tease out causation.

Another obvious point worth mentioning is that many “drug externalities” are actually caused by prohibition. Gang violence, heroin overdoses, HIV epidemics, IV drug use *in general*, economic crimes committed to acquire drugs (because the market price has been raised, an actual *intentional* effect of prohibition), and many other drug-related social harm simply do not exist in a regime of legalization. Or perhaps they exist, but are smaller, and they occur in the light of day where they can be managed. 

Market Wages are Fair

Those really high and really low market wages, the ones everyone gets upset about, are pretty much fair.

Due to economies of scale, sometimes it makes sense for a project, factory, business, investment fund, or other asset to be very large. Say in the $1 billion range. It usually makes sense for there to be an executive decision-maker. Otherwise the massive (and massively valuable) business or project has no central brain and can be pulled in multiple conflicting directions. A bad executive can destroy a significant fraction of the business’s or project’s value. So it probably makes sense to pay an executive tens of millions of dollars for these large projects to avoid such a fate. There will certainly be a very large difference between the value-added of a typical executive and that of an average worker pulled at random from the population, so right away it makes sense that these folks would be rewarded for their special talents. It may even be true that the difference between the very best executive and her closest runner-up is enormous. Suppose the very best executive has a 1% chance of completely destroying the full value of a $1 billion firm (or asset or project), while his closest rival has a 2% chance of wreaking the same destruction. The best executive’s value added is in the $10 million range on this consideration alone. It probably makes sense to pay tens or even hundreds of millions of dollars to people who make billion dollar decisions.

Given all this, the decision to shell out for top executive talent is probably a “fair” market outcome. It’s also an efficient one that’s good for all stakeholders. Workers benefit from being hitched to a well-managed project or company rather than a poorly managed one. They are less likely to find themselves suddenly thrown out of work if the project or company fails, and their overall productivity is higher. Shareholders likewise benefit from higher returns on their investment. And customers likewise benefit from buying a better product (higher in quality, or the same quality for cheaper due to better efficiencies, or never-before-existing, or whatever “better” means in your book). It’s not terribly surprising that shareholders, spending large sums of their own money on executive compensation, often pay out multi-million dollar compensation packages. Paraphrasing Thomas Sowell here, it makes little sense to be penny-wise and pound-foolish when billion-dollar decisions are being made. It simply isn’t true, as is often remarked, that the very high compensation paid to executives is *at the expense of* these other stakeholders. It makes far more sense to think of executive pay as a wise investment.

Those very low market wages are also probably fair, sorry to say. Wages reflect productivity, and low-wage workers just don’t add much value for all their toil. Many believe that employers hold “market power” and have managed to artificially keep wages low for some workers. But this story doesn’t make much sense when you apply even a tiny amount of scrutiny to it. Why only some workers and not others? Why do firms that hire highly-skilled workers bother to pay them more than the low-skilled workers? If one company is reaping a big profit by underpaying its low-skilled workers, why doesn’t another firm come along that reaps slightly less profit by paying all those workers just slightly more (and then another one offering to accept slightly less profit by offering higher wages, and so on until the excess profit has been competed away)? Many low-wage workers impose costs on their employers beyond their wages + benefits. Think an impulsive teenager or an ex-convict working in a customer-facing job (like a cashier). A bad attitude from a single cashier could lose the store a customer for life, and all future revenue from that customer. An employer likely won’t take a chance on such a person unless they can hire him at a very low wage.

There is a lot of magical thinking around wages. Most of it is populist drivel. There are underlying reasons for the patterns you see in a market economy. Market transactions are voluntary in the sense that they don’t occur unless both parties agree to the terms. So there is probably a good reason why people do as they do. If you find a certain market outcome distasteful, you should at least try to understand the underlying reasons. We can’t magically raise or lower wages by sheer force of will. And we can’t legislate a distribution of wages different from the one that arises naturally without causing a great deal of harm.

Workers for the most part earn a wage that is commensurate with the value of what they produce. If ever this *weren’t* the case, market forces would push wages in the appropriate direction. You don’t have to appeal to some obscure theory of economics to understand why this is true. Logic alone should tell you that if someone is underpaying for something, another bidder will come along to bid up the price. And if someone is *overpaying* for something, they will either realize their folly and stop or lose out to a party who is wiser about managing its expenses.

There are no doubt market distortions. Bad government policies can, say, distort the optimum firm size so that there are more mega-firms and thus higher paid executives. Or they can distort the mix of pay and benefits for low-wage workers such that they receive less than optimum market wages. I’ll join any call to repeal such government interference in the marketplace, but there would no doubt still be a mix of very high and very low wages. The overall distribution would still be very uneven, probably close to what it is now. While some poorly conceived government policies might increase income inequality, it’s a mistake to label the inequality itself as unfair. Do so and you fixate on the wrong problem and implement the wrong solution.